1# mach: crisv3 crisv8 crisv10 crisv32 2# output: ffffff00\n 3 4; A bug resulting in a non-effectual clear.b discovered running the GCC 5; testsuite; jump actually wrote to p0. 6 7 .include "testutils.inc" 8 9 start 10 jump 1f 11 nop 12 .p2align 8 131: 14 move.d y,r4 15 16 .if 0 ;0 == ..asm.arch.cris.v32 17; There was a bug causing this insn to set special register p0 18; (byte-clear) to 8 (low 8 bits of location after insn). 19 jump [r4+] 20 .endif 21 221: 23 move.d 0f,r4 24 25; The corresponding bug would cause this insn too, to set p0. 26 jump r4 27 nop 28 quit 290: 30 moveq -1,r3 31 clear.b r3 32 checkr3 ffffff00 33 quit 34 35y: 36 .dword 1b 37